Hybridization Denoising Method for EMG Signals Using EWT and EMD Techniques

Abstract

Electromyography (EMG) is a key diagnostic tool for assessing muscle electrical activity, aiding in the diagnosis of neuromuscular disorders, and evaluating muscle health. However, EMG signals are often marred by noise, making accurate diagnostics challenging. This study introduces a hybrid denoising method that combines Empirical Wavelet Transform (EWT) and Empirical Mode Decomposition (EMD). By effectively separating desired signals from noise, this method improves signal analysis accuracy by leveraging the advantages of both techniques. The EMD, EWT, and the hybrid EWT-EMD methods have been tested on MIT-BIH databases containing real EMG signals with varying levels of composite noise. The results have demonstrated that the hybrid method not only excels in noise reduction but also maintains the integrity of the original EMG signal, which is crucial for accurate diagnosis and analysis. By comparing metrics such as Signal-to-Noise Ratio (SNRout) and Mean Squared Error (MSE, PRD, and RMSE) across the different methods, the results have confirmed that the hybrid approach has consistently outperformed the individual EMD and EWT methods on normal signals, and the hybrid method and EWT on abnormal signals. This hybrid approach holds great potential for improving diagnostic accuracy and patient care in neuromuscular research and clinical practice, ultimately contributing to better healthcare outcomes.